Bio-Orthogonal Chemistry Conjugation Strategy Facilitates Investigation of N-methyladenosine and Thiouridine Guide RNA Modifications on CRISPR Activity

Abstract

The CRISPR-Cas9 system is an important genome editing tool that holds enormous potential toward the treatment of human genetic diseases. Clinical success of CRISPR technology is dependent on the incorporation of modifications into the single-guide RNA (sgRNA). However, chemical synthesis of modified sgRNAs, which are over 100 nucleotides in length, is difficult and low-yielding. We developed a conjugation strategy that utilized bio-orthogonal chemistry to efficiently assemble functional sgRNAs containing nucleobase modifications. The described approach entails the chemical synthesis of two shorter RNA oligonucleotides: a 31-mer containing tetrazine (Tz) group and a 70-mer modified with a trans-cyclooctene (TCO) moiety. The two oligonucleotides were conjugated to form functional sgRNAs. The two-component conjugation methodology was utilized to synthesize a library of sgRNAs containing nucleobase modifications such as N¹-methyladenosine (m¹A), N⁶-methyladenosine (m⁶A), 2-thiouridine (s²U), and 4-thiouridine (s⁴U). The impact of these RNA modifications on overall CRISPR activity were investigated in vitro and in Cas9-expressing HEK293T cells.

Introduction

The CRISPR-Cas9 system is a powerful genome editing tool that profoundly revolutionized biomedical research and holds enormous potential for successful treatment of human genetic diseases. The two-component system, consisting of a single-guide RNA (sgRNA) and a DNA nuclease, Cas9, forms a complex that is capable of targeting double-stranded DNA with single nucleotide precision.^1,2 This technology has been successfully applied for genome editing at the cellular level,^3,4 as well as in vivo.^5–7 In recent years, CRISPR technology has been evaluated in several human clinical trials for the treatment of blood disorders,⁸ various cancers,^9–11 eye disease,¹² and chronic infection.¹³

Virtually all FDA-approved RNA-based therapeutics contain RNA modifications.^14–17 The clinical success of CRISPR technology is likely dependent on well-designed incorporation of RNA modifications that can potentially improve sgRNA stability, on-target efficiency and lower off-target effects. A number of RNA modifications have already been explored to achieve these goals.^18–21 None of the reported systems, however, examined sgRNA containing nucleobase modifications, which are common in nature and capable of regulating critical cellular processes.²²

This is likely due to the fact that sgRNA is over 100-nt long, and solid phase synthesis (SPS) of long oligonucleotides is difficult. Further, the incorporation of nucleobase-modified RNA phosphoramidites during SPS exacerbates the problem, as it is typically associated with lower coupling yields. Our goal has been to develop a versatile synthetic approach that will allow to examine the effects of modified nucleobases such as N¹-methyladenosine (m¹A), N⁶-methyladenosine (m⁶A), 2-thiouridine (s²U), and 4-thiouridine (s⁴U) on CRISPR activity.

We were inspired by the recently reported “split-and-click” strategy, illustrated in Figure 1A, which utilizes copper-catalyzed azide alkyne cycloaddition (CuAAC) chemistry to form functional sgRNAs from two smaller oligonucleotide fragments.²³ This approach entails modification of the 3′-end of RNA 1 oligonucleotide with an alkyne group and modification of the 5′-end of RNA 2 oligonucleotide with an azide group. The two reacting groups are small, and the resulting triazole linker has been shown to cause minimal structural perturbation to “clicked” sgRNAs.

FIG. 1.

(A) Previously reported “split-and-click” approach to form functional sgRNAs using CuAAC chemistry. (B) Synthesis of conjugated sgRNAs using IEDDA chemistry. CuAAC, copper-catalyzed azide alkyne cycloaddition; IEDDA, inverse electron demand Diels-Alder; sgRNA, single-guide RNA.

One of the drawbacks of the reported approach is that the ligation chemistry is catalyzed by Cu(I), which is biologically toxic and should ideally be avoided in the synthesis of therapeutic RNAs.²⁴ Our goal was to investigate whether the “split-and-click” strategy could be successfully accomplished using inverse electron demand Diels-Alder (IEDDA) chemistry, which does not require Cu(I) catalysis. The bio-orthogonal IEDDA chemistry has been shown to be compatible with the intricate functional groups and structural elements of RNA.^25–27

As illustrated in Figure 1B, our approach entails modification of the 2′-end of RNA 1 oligonucleotide with a tetrazine (Tz) group and modification of the 5′-end of RNA 2 oligonucleotide with a trans-cyclooctene (TCO) moiety. The two oligonucleotides can be conjugated in a biological buffer under neutral pH and without any transition metal catalyst, thus forming “conjugated” sgRNA (Fig. 1B). Watts and co-workers recently reported a similar bioconjugation strategy bridging two segments of sgRNA in the tetraloop region using Tz and norbornene.²⁸

Materials and Methods

All oligonucleotide solid phase syntheses were done on a 1.0 μmol scale using the Oligo-800 synthesizer (Azco Biotech, Oceanside, CA). Solid phase syntheses were performed on controlled pore glass. (CPG-1000) purchased from Glen Research (Sterling, VA). Other oligonucleotide SPS reagents were obtained from ChemGenes Corporation (Wilmington, MA). Phosphoramidites (TBDMS as the 2′-OH protecting group): rA was N-Bz protected, rC was N-Ac protected, and rG was N-iBu protected. A, C, G, and U phosphoramidites were dissolved in anhydrous acetonitrile (0.07 M) directly before use. m¹A, m⁶A, s²U, and s⁴U phosphoramidites were dissolved in anhydrous acetonitrile (0.15 M) directly before use.

A coupling step was done using 5-ethylthio-1H-tetrazole solution (0.25 M) in acetonitrile for 12 min. 5′-Detritylation step was done using 3% trichloroacetic acid in CH₂Cl₂. Oxidation step was done using I₂ (0.02 M) in THF/pyridine/H₂O solution. The CPG modifications were carried out using native amino lcaa CPG 1000 A˚, purchased from ChemGenes Corporation (Cat. No. N-5100-10).

For gel electrophoresis, 10 × Tris/Borate/EDTA (TBE) buffer was purchased from Fisher Scientific Company, LLC (Waltham, MA) and used with proper dilution. Thirty percent Arcylamide/Bis-arcylamide solution (29:1) was purchased from Bio-Rad Laboratories, Inc. (Hercules, CA). Chromatographic purifications of synthetic materials were conducted using SiliaSphere^™ spherical silica gel with an average particle and a pore size of 5 μm and 60 A˚, respectively (Silicycle, Inc., Quebec, Canada). Thin layer chromatography (TLC) was performed on SiliaPlate^™ silica gel TLC plates with 250 μm thickness (Silicycle, Inc.).

Flash chromatography was performed using Biotage Isolara One instrument (Biotage Sweden AB, Uppsala, Sweden). Preparative TLC was performed using SiliaPlate silica gel TLC plates with 1000 μm thickness. ¹H, ¹³C, and ³¹P nuclear magnetic resonance (NMR) spectroscopy was performed on a Bruker NMR at 500 MHz (¹H) and 126 MHz (¹³C) and 121 MHz (³¹P). All ¹³C NMR spectra were proton decoupled. High-resolution electrospray ionization–mass spectrometry (ESI-MS) spectra of small molecules was acquired using Agilent Technologies 6530 Q-TOF instrument. The ESI-MS analysis of RNA oligonucleotides was carried out by Novatia, LLC (Newtown, PA).

Synthesis of RNA oligonucleotides containing TCO

The phosphoramidite 7, containing TCO, was coupled during the final synthetic cycle of SPS of RNA. The subsequent oxidation step was performed using 1 M solution of t-BuOOH in CH₂Cl₂. After completion of SPS, CPG beads were treated with concentrated aqueous ammonia in a sealed vial for 2 h at 55°C. The ammonia was removed in vacuo (SpeedVac). Cleaved oligonucleotides were redissolved in mixture of anhydrous DMSO (100 μL) and triethylamine trihydrofluoride (125 μL) and heated for 2.5 h at 65°C.

After subsequent cooling to room temperature (rt), sodium acetate buffer (3 M, pH 5.2, 25 μL) and ethanol (1 mL) were added, and the RNA was stored overnight at −20°C. The RNA was then pelleted by centrifugation (14,000 g, 17 min, 4°C), the supernatant was discarded, and the pellet was washed with 75% ethanol (1 mL). The pellet was then dried in vacuo, and it was dissolved in water.

Assessment of TCO coupling to RNA oligonucleotides

The CPG beads containing RNA 2 oligo-TCO (5 mg) were mixed with Tz-DMT for 1 h at rt. Synthesis of Tz-DMT was previously described (He et al²⁵). After the reaction, the beads were thoroughly washed with DMF and CH₂Cl₂. The CPG beads were treated with 10 mL deblocking solution (3% trichloroacetic acid in CH₂Cl₂). Absorbance at 504 nm was used to calculate the amount of cleaved DMT group (ɛ = 76 mL·cm⁻¹·μmol⁻¹). The amount of coupled TCO was calculated based on the assumption of nearly quantitative reaction between RNA 2 oligo-TCO and Tz-DMT.

Purification of RNA oligonucleotides by preparative polyacrylamide gel electrophoresis

Oligonucleotides were mixed with formamide dye (50% v/v) and loaded onto a denaturing 10–20% polyacrylamide gel (1 × TBE buffer containing 8 M urea) and separated at 500 V for 1–2 h. The RNA bands were visualized under ultraviolet, excised, crushed, and soaked in buffer [ammonium acetate 500 mM, Mg(OAc)₂ 10 mM, EDTA 2 mM) 800 μL] for 16 h at rt with vigorous shaking. Buffer volume was reduced to 50 μL by butanol extraction (5 × 500 μL). Cold ethanol (1000 μL) was added, and RNA was precipitated at −20°C overnight.

Cas9 in vitro cleavage assay

pBR322 plasmid DNA (0.35 μM, 1.13 μL, N3033S; NEB) was diluted with water (16.87 μL) and NEB buffer 3.1 (10 × , 2 μL). The plasmid was linearized directly before CRISPR experiments using PvuII (10 U/μL, 1 μL, R0151S; NEB) for 1 h at 37°C. In analogous fashion, eGFP-N1 plasmid was linearized with DraIII-HF (10 U/μL, 1 μL, R3510L; NEB). For the Cas9-mediated DNA cleavage assay, sgRNA (300 nM, 5 μL), Cas9 (1 μM, 0.3 μL, M0386S; NEB), Cas9 buffer (10 × , 1 μL; NEB), linearized plasmid (20 nM, 1.5 μL), and H₂O (2.2 μL) were mixed together (final volume = 10 μL) and incubated for either 1 or 16 h at 37°C.

CRISPR experiments were terminated by the addition of proteinase K (20 mg/mL, 0.5 μL) for 1 h at 37°C. The reaction (1 μL) was mixed with blue loading buffer (6 × , 2 μL, B7703S; NEB) and loaded on a 1% agarose stained with ethidium bromide (1 × TBE running buffer).

Evaluation of in-cell CRISPR activity

CRISPR-Cas9 experiments were carried out following the procedure reported by Yin et al.²⁰ HEK293T cells were infected by lentiviral particles to stably express EF1a-GFP-PGK-Puro (26777; Addgene) and EFs-spCas9-Blast (52962; Addgene). The cells were grown to 70–90% confluence in Dulbecco's modified Eagle's medium (DMEM), containing 10% fetal bovine serum (FBS). The cells were transfected with GFP-targeting sgRNAs (30 nM each, final concentration) using Lipofectamine (Thermo Fisher Scientific) for 24 h. The cells were further grown in DMEM for 48 h post-transfection.

The cells were fixed with 2% paraformaldehyde in PBS, and GFP expression was analyzed by flow cytometry. Data from 10⁶ cells were acquired using a FACS Aria III cell sorter equipped with a 488 nm/blue coherent sapphire solid-state laser, 20 mW (BD Biosciences, San Jose, CA). Data analyses were carried out using FlowJo software (Ashland, OR), according to the manufacturer's instructions. Parameters, such as median fluorescence intensity (MFI) and the percentages of specific populations, were quantified by histogram analysis.

Synthesis of conjugated sgRNA using IEDDA chemistry

RNA 1 oligo 1 containing Tz (250 μmol) and RNA 2 oligo 2 containing TCO (120 μmol) were mixed in an NaCl solution (0.2 M, 7.9 μL), containing EDTA (0.5 M, 0.2 μL) and HEPES (0.2 M, pH 7.5, 2 μL). The reaction mixture was agitated at 1000 rpm at rt for 16 h. The conjugated sgRNA was purified by preparative PAGE.

Results

The synthesis of the two RNA components described in our strategy for making conjugated sgRNA is illustrated in Figure 2. Toward attachment of Tz to the 2′-end of RNA 1 oligonucleotide, we synthesized controlled pore glass (CPG) solid support 3 that was modified with a uridine analog containing trifluoroacetyl-protected amine group at the 2′-position.^29,30 After completion of the SPS of RNA and subsequent cleavage, deprotection, and desilylation steps, the resulting RNA 1 oligonucleotide will contain an amine group that was coupled to the Tz-NHS ester 5. Our choice of Tz was based on the reports that indicated its exceptional stability under physiological conditions.^31,32

FIG. 2.

(A) Synthesis of the RNA oligonucleotide-modified with Tz; (B) Synthesis of the RNA oligonucleotide-modified with TCO. TCO, trans-cyclooctene; Tz, tetrazine.

The Tz-modified RNA oligonucleotides were purified by preparative PAGE and characterized by ESI-MS (Supplementary Figs. S1–S14). As illustrated in Figure 2B, 5′-end of RNA 2 oligonucleotide was modified with TCO using previously reported TCO-phosphoramidite 7.³³ Compound 7 was incorporated at the final coupling step of SPS. Modified oxidation step (Supplementary Data) using tert-butyl hydroperoxide was used due to sensitivity of TCO to the standard SPS oxidation chemistry. The TCO-modified RNA oligonucleotides were obtained after the standard cleavage, deprotection, and desilylation steps.

FIG. 5.

(A) Activity of unmodified sgRNA and sgRNA 4, sgRNA 5, and sgRNA 6 constructs in facilitating Cas9-mediated cleavage of linearized eGFP-N1 plasmid at 1 and 16 h time points. All experiments were done in duplicate. (B) Activity of sgRNAs containing m¹A, m⁶A, s²U, and s⁴U modifications (sgRNA 7, sgRNA 8, sgRNA 9, and sgRNA 10) in facilitating Cas9-mediated cleavage of linearized eGFP-N1 plasmid at 1 and 16 h time points. (C) Activity of sgRNAs containing one, two, and three s²U and s⁴U modifications (sgRNA 9, sgRNA 10, sgRNA 11, sgRNA 12, sgRNA 13, and sgRNA 14) in facilitating Cas9-mediated cleavage of linearized eGFP-N1 plasmid at 1 and 16 h time points. (D) Summary of Cas9-mediated cleavage of linearized eGFP-N1 plasmid after 16 h. m¹A, N¹-methyladenosine; m⁶A, N⁶-methyladenosine; s²U, 2-thiouridine; s⁴U, 4-thiouridine.

The conjugated sgRNAs that are formed via IEDDA chemistry contain a linker that is considerably larger than the triazole linker (Fig. 1A). According to experimental analysis described by Taemaitree et al, this could potentially be detrimental for CRISPR activity due to perturbation of native interactions between sgRNA and Cas9.²³ Our goal was to find a site where longer linkers would be well tolerated. We were inspired by Iwasaki et al, who reported sgRNA constructs containing fused theophylline aptamer capable of regulating CRISPR activity.³⁴

We synthesized three sgRNAs in which the two RNA segments were conjugated together at different positions in the repeat region, as shown in Figure 3. The conjugated sgRNAs were purified by preparative PAGE, as illustrated in Supplementary Figure S15. The sequence of the constructed sgRNAs has been reported to target linearized pBR322 plasmid.²³

FIG. 3.

Synthesis of conjugated sgRNA constructs by ligating two RNA fragments using IEDDA chemistry.

The constructs described in Figure 3 were tested in vitro for their ability to carry out Cas9-mediated cleavage of linearized pBR322 plasmid. All of the CRISPR experiments described here and beyond were carried out in duplicate. As illustrated in Figure 4, sgRNA 1 and sgRNA 2 were unable to facilitate DNA cleavage, presumably due to the prohibitively long linker. Fortunately, sgRNA 3 was found to be active (lanes 6 and 7). In fact, sgRNA 3 outperformed the unmodified sgRNA (lanes 2 and 3).

FIG. 4.

Activity of unmodified sgRNA, sgRNA 1, sgRNA 2, and sgRNA 3 constructs in facilitating Cas9-mediated cleavage of linearized pBR322 plasmid at 1 and 16 h time points.

These findings proved that placement of the long linker is crucial to facilitate proper RNA-Cas9 interactions that enable nuclease activity. To test the generality of our approach, we synthesized sgRNA 4 targeting the GFP gene encoded on the linearized eGFP-N1 plasmid (Table 1 and Supplementary Table S1). The new construct was linked at the same location in the repeat sequence as sgRNA 3. As illustrated in Figure 5A, sgRNA 4 (lanes 4 and 5) was also found to be active also outperforming the corresponding unmodified sgRNA (lanes 2 and 3).

Table 1.

Sequences of the CRISPR RNA region of GFP-targeting single guide RNAs containing nucleobase modifications

Construct	Sequence	Note
sgRNA 4	5′-GGG CGA GGA GCU GUU CAC CGG UUU …-3′	Unmodified RNA
sgRNA 5	5′-GGG CGA GGA GCA GUU CAC CGG UUU …-3′	One U to A mismatch
sgRNA 6	5′-GGG CGA GGA GCA GAA CAC CGG UUU …-3′	Three U to A mismatches
sgRNA 7	5′-GGG CGm1A GGm1A GCU GUU Cm1AC CGG UUU …-3′	Three m1A modifications
sgRNA 8	5′-GGG CGm6A GGm6A GCU GUU Cm6AC CGG UUU …-3′	Three m6A modifications
sgRNA 9	5′-GGG CGA GGA GCs2 UGs2 Us2U CAC CGG UUU …-3′	Three s2U modifications
sgRNA 10	5′-GGG CGA GGA GCs4UGs4Us4U CAC CGG UUU …-3′	Three s4U modifications
sgRNA 11	5′-GGG CGA GGA GCs2U GUU CAC CGG UUU …-3′	One s2U modification
sgRNA 12	5′-GGG CGA GGA GCs2UGs2U U CAC CGG UUU …-3′	Two s2U modifications
sgRNA 13	5′-GGG CGA GGA GCs4U GUU CAC CGG UUU …-3′	One s4U modification
sgRNA 14	5′-GGG CGA GGA GCs4UGs4U U CAC CGG UUU …-3′	Two s4U modifications

m¹A, N¹-methyladenosine; m⁶A, N⁶-methyladenosine; s²U, 2-thiouridine; s⁴U, 4-thiouridine; sgRNA, single-guide RNA.

Empowered by the synthetic methodology to assemble functional conjugated sgRNAs, we turned our attention toward investigation of the effects of nucleobase modifications m¹A, m⁶A, s²U, and s⁴U on CRISPR activity. The chosen RNA modifications are known to play important regulatory roles in cellular biochemistry. m¹A has been found in ribosomal RNA (rRNA), transfer RNA (tRNA), and messenger RNA (mRNA).^35–38 When incorporated into mRNA, m¹A can upregulate translation.³⁶ A single m¹A modification in the vicinity of the start codon resulted in a 1.2–1.4 higher mRNA translation efficiency in both human and mouse cells.³⁶

Similarly, recent transcriptome-wide m⁶A-seq studies also discovered m⁶A RNA modification in 5′-untranslated region (UTR) and start codons.^39,40 m⁶A regulates translation of mRNA by recruiting m⁶A reader proteins, such as YTHDF1, YTHDF2, and YTHDF3; as well as YTHDC1 and YTHDC2,⁴¹ which promote mRNA translation. Another recent study reported that m⁶A also repels certain RNA-binding proteins, forming an additional layer of regulation of mRNA translation.⁴² Meanwhile, s²U and s⁴U are known to be present in certain tRNAs. s²U has been found at position 34 of tRNA, the first position of the anticodon, which is base-paired with the nucleotide at the wobble position of the mRNA codon.⁴³

The presence of s²U has been attributed to codon-anticodon recognition efficiency and accuracy. The RNA modification has been shown to enhance the aminoacylation kinetics of tRNA, and to prevent frame-shifting during translation.⁴⁴

The aforementioned RNA modifications were incorporated in the CRISPR RNA (crRNA) region (Table 1), which is responsible for the recognition of complementary target DNA.^1,2 Based on previously reported studies, m¹A, m⁶A, and s⁴U will destabilize RNA-DNA duplex, whereas s²U will have stabilizing effects.^45–47 Our convergent approach facilitated efficient assembly of sgRNAs 5–10 using the two-component conjugation strategy. The RNA modifications, whose coupling is often associated with lower yields, were incorporated into the shorter, 31-nt long, RNA 1 oligonucleotide (Fig. 1B). sgRNAs 5–10 share the same 70-nt long RNA 2 oligonucleotide component (Fig. 1B).

First, to get an idea of how many modifications would be necessary to observe a pronounced effect, we synthesized sgRNA 5 and sgRNA 6 (Table 1) that contain either one or three U to A mismatches. As illustrated in Figure 5A, a single mismatch in sgRNA 5 does not have a pronounced impact on the nuclease activity. On the other hand, the incorporation of three U to A mismatches in the sgRNA 6 rendered it completely inactive. Therefore, to investigate the impacts of RNA modifications, we decided to synthesize sgRNAs 7, 8, 9, and 10 containing three m¹A, m⁶A, s²U, and s⁴U modifications, respectively (Table 1).

As illustrated in Figure 5B, the incorporation of three m¹A modifications in sgRNA 7 completely eliminated nuclease activity. This was an expected outcome, as methylation at the 1-position of adenosine disrupts hydrogen bonding to a complimentary thymidine. Consequently, three m¹A residues disrupted the ability of sgRNA 7 to recognize and bind the complimentary double-stranded DNA (dsDNA). Analogous behavior was observed with sgRNA 10, which contains three s⁴U residues. S⁴U was reported to destabilize the duplex by 0.6 kcal/mol, which was attributed to weaker Watson–Crick base pairing due to the replacement of the hydrogen bond accepting oxygen by a weaker hydrogen bond accepting sulfur at C4.⁴⁵

Meanwhile, quite unexpected results were observed with sgRNA 8 and sgRNA 9. It has been reported that s²U modification leads to increased duplex stability by about 1 kcal/mol.^46,48 The increased stability has been attributed to enhanced stacking interactions and higher acidity of the N-3 proton and thus increased strength of its hydrogen bond.^46,48 However, the incorporation of three s²U residues in sgRNA 9 led to lower CRISPR activity relative to sgRNA 4, which contains canonical nucleobases. On the other hand, sgRNA 8, containing three m⁶A residues exhibited higher nuclease activity than sgRNA 4.

This finding was also unexpected, as N⁶-methylation is known to have a destabilizing effect in base-paired regions of RNA by 0.5–1.7 kcal/mol. The N⁶-methyl group has been predicted to be in a high-energy anti conformation, with the methyl group oriented into the major groove of the RNA, where the amino proton would normally be.⁴⁷

Hoping to find some predictable trends, we decided to further investigate the impacts of s²U and s⁴U nucleobase modifications on CRISPR activity. We synthesized sgRNA 11 and sgRNA 12 that contain one and two s²U modifications, as well as sgRNA 13 and sgRNA 14 that contain one and two s⁴U modifications (Table 1 and Supplementary Table S1). The new constructs were tested for their ability to carry out Cas9-mediated cleavage of linearized eGFP-N1 plasmid and correlated to sgRNA 9 and sgRNA 10 that contain three s²U and s⁴U modifications, respectively. As illustrated in Figure 5C, all of sgRNAs containing s²U RNA modifications exhibited lower nuclease activity than the unmodified sgRNA 4, without any noticeable trends.

The sgRNA 14, containing two s⁴U modification, was found to be completely inactive, just like the previously tested sgRNA 10. Perhaps, the most interesting result was observed with sgRNA 13, which contains a single s⁴U modification. It was found to have the highest nuclease activity of all the tested modified sgRNAs. All of the described CRISPR experiments are summarized in Figure 5D.

The ability of conjugated sgRNAs, containing m⁶A and s²U modifications, to enable Cas9 nuclease activity was tested in GFP and Cas9-expressing HEK293T cells.²⁰ The cells were transfected for 24 h with the GFP-targeting sgRNA 4, sgRNA 8, and sgRNA 9, as well as unmodified sgRNA, as a positive control. The cells were analyzed by flow cytometry 48 h post transfection to investigate the expression of GFP. As illustrated in Figure 6, the untransfected cells show two sub-populations of GFP-expressing cells.

FIG. 6.

Flow cytometry analysis of GFP expression of GFP and Cas9-expressing HEK293T transfected with native sgRNA and sgRNA4.

The subpopulation with higher GFP fluorescence corresponds to 4.16% ± 0.44% of total cells, whereas the subpopulation with lower GFP signal corresponds to 78.93% ± 0.25% of total cells. Transfection with unmodified sgRNA almost completely eliminated the subpopulation of cells with higher GFP fluorescence (0.25% ± 0.14%), while also decreasing the subpopulation of cells with a lower GFP signal to 66.37% ± 3.69%. Analogous behavior was observed with the cells transfected with sgRNA 4 (higher GFP subpopulation 0.20% ± 0.05%; lower GFP subpopulation 67.07% ± 1.18%).

This indicated that sgRNA4 is functional inside the cells to enable nuclease activity that is on par with the native system. Contrary to the prediction from the in vitro experiments, sgRNA 8 and sgRNA 9 were found to have lower CRISPR activity inside the cells. The flow cytometry analysis illustrated in Figure 7 indicates that the cells transfected with sgRNA 8 had a higher GFP-expressing subpopulation of 2.12% ± 0.07% and a lower GFP-expressing subpopulation of 74.07% ± 0.60%. Similarly, the cells transfected with sgRNA 9 had a higher GFP-expressing subpopulation of 2.31% ± 0.11% and a lower GFP-expressing subpopulation of 69.80% ± 0.85%. These data are summarized in Figure 8.

FIG. 7.

Flow cytometry analysis of GFP expression of GFP and Cas9-expressing HEK293T transfected with sgRNA8 and sgRNA9.

FIG. 8.

Flow cytometry analysis of GFP expression of GFP and Cas9-expressing HEK293T transfected with unmodified sgRNA, sgRNA 4, sgRNA8, and sgRNA9. (A) Subpopulation of cells with high GFP fluorescence, (B) subpopulation of cells with medium GFP fluorescence. All flow cytometry experiments were done in triplicate.

Discussion

The CRISPR-Cas9 system is a powerful genome editing tool whose clinical translation will likely require thorough structural optimization. Cas9-guided nuclease activity is dependent on the interactions between the enzyme and sgRNA, as well as the subsequent recognition and association between this Cas9-sgRNA complex and target dsDNA. This multi-step process has been demonstrated by structural studies to involve a cascade of conformational changes within the RNA-enzyme-DNA system.^49,50 In recent years, a number of different strategies have evolved to manipulate and optimize each of the three components and improve the overall CRISPR efficiency and specificity of DNA cleavage.

For example, new and improved Cas9 variants have been reported, such as eSpCas9, SpCas9-HF1, and HypaCas9.^51–53 In comparison, the strategy of changing the sequence, stability or conformation of guide RNA requires relatively less effort than new enzyme engineering. Riesenberg et al incorporated highly stable hairpins into guide RNA to ensure proper folding that leads to improved genome editing regardless of spacer sequence composition.⁵⁴ It has been shown that chemical modifications on guide RNA such as 2′-deoxy, 2′-F, 2′-O-methyl, locked nucleic acids (LNA), and phosphorothioated backbone can also improve nuclease activity and gene editing efficiency probably due to both enhanced interactions within the CRISPR system and increased biostability of RNA.^18,55–58

However, balancing and optimizing the Cas9 cleavage efficiency and on-target specificity remains a big challenge for gene editing and requires more systematic investigation. From this perspective, using naturally or artificially modified nucleobases to further adjust the interactions of crRNA, Cas9 and target DNA represents a promising approach. There are many base modifications that are currently available to fine-tune the stability and specificity of RNA binding with both Cas9 and DNA. Due to the intrinsic synthetic challenges of making long base-modified RNA strands, such methods have not been well explored.

Our metal-free IEDDA chemistry-based bio-orthogonal method improves the RNA synthesis yield and facilitates construction of a model platform to thoroughly study the overall DNA cleavage effects of all the base modifications with available phosphoramidite building blocks.

In the current work, we demonstrated that the TCO-Tz linkages can be compatible within the CRISPR system if placed in the optimized position of sgRNA. The developed conjugation strategy can potentially be employed for a variety of research application. It can be used to facilitate fluorescent labeling or tagging of sgRNAs. It can also be utilized to create sgRNA libraries.²³ In our work, the bio-orthogonal conjugation strategy has been explored to examine impacts of RNA nucleobase modifications on CRISPR efficiency.

We selected several naturally occurring RNA modifications, m¹A, m⁶A, s⁴U, and s²U, which are yet to be examined in the context of CRISPR experiments. Since m¹A and m⁶A destabilize RNA-DNA duplex, whereas s²U and s⁴U have stabilizing effects, we set out to test correlation with the overall CRISPR efficiency.

It turns out that three s⁴U or m¹A residues completely inhibit the DNA cleavage activity probably due to the resulting unstable crRNA-DNA duplex. In comparison, three m⁶A significantly increase the cleavage activity, which is contradictory to its destabilization effect on RNA-DNA. Similarly, three s²U residues, which are expected to largely enhance the duplex stability, have a slightly negative impact on the cleavage activity. Of the 14 surveyed sgRNAs, the highest nuclease activity was exhibited by the construct containing a single s⁴U modification, which was also quite unexpected.

These results indicate that there might be no direct correlations between the stabilization effect and CRISPR activity. Although more systematic structural and biophysical insights are required to investigate the impacts of these modifications on Cas9 interactions, most likely the enzyme recognition of crRNA has been changed with these modifications. In addition, for the modifications such as m⁶A that can increase the DNA cleavage efficiency, the cleavage specificity or off-target effect remains to be investigated. Nonetheless, this work provides a useful platform to test more chemical modifications and identify some useful candidates to further increase both CRISPR efficiency and precision.

Authors' Contributions

A.H. carried out chemical synthesis of RNA phosphoramidites and SPS of RNA containing m¹A and m⁶A modifications. A.H. performed in vitro CRISPR experiments. Y.Y.Z. carried out SPS of RNA containing s²U and s⁴U modifications. Y.Y.Z. performed in-cell CRISPR experiments and flow cytometry. M.R. and J.S. conceptualized the project, and they provided student supervision and funding for the project. M.R. and J.S. wrote the manuscript with input from all authors. All authors approved the final manuscript.

Author Disclosure Statement

There are no conflicts to declare.

Funding Information

The authors gratefully acknowledge financial support by the National Institutes of Health of the United States (Grant R21HG012257 to M.R.) and the National Science Foundation (Grant CHE-1664577 to M.R.; Grants CHE-1845486 to J.S.). This work was also supported, in part, by the National Institutes of Health of the United States Grant R15GM12811501.

Supplementary Material

Supplementary Data

Supplementary Figure S1

Supplementary Figure S2

Supplementary Figure S3

Supplementary Figure S4

Supplementary Figure S5

Supplementary Figure S6

Supplementary Figure S7

Supplementary Figure S8

Supplementary Figure S9

Supplementary Figure S10

Supplementary Figure S11

Supplementary Figure S12

Supplementary Figure S13

Supplementary Figure S14

Supplementary Figure S15

Supplementary Table S1